点击显示 收起
1 From the Medical Research Council Childhood Nutrition Research Center (AS, RM, KK,EI, MF, and AL), the Neurodevelopmental & Neurodisability Unit, Department of Neurology, The Wolfson Center (PS), and the Centre for Paediatric Epidemiology and Biostatistics (TJC), Institute of Child Health, London, United Kingdom; the Department of Pediatrics and the Murdoch Children's Research Institute, Royal Children's Hospital, Melbourne, Australia (RM); Leicester General Hospital NHS Trust, Leicester, United Kingdom (AE-J); and the Faculty of Medicine and Health Sciences, Academic Division of Child Health, University of Nottingham, Nottingham, United Kingdom (TS)
2 Supported by the Medical Research Council, United Kingdom. Nestec Ltd (Vevey, Switzerland) donated the trial formulas. 3 Address reprint requests to A Singhal, Medical Research Council Childhood Nutrition Research Center, Institute of Child Health, 30 Guildford Street, London, WC1N 1EH, United Kingdom. E-mail: a.singhal{at}ich.ucl.ac.uk
ABSTRACT
Background: Breastfeeding has been reported to benefit visual development in children. A higher concentration of docosahexaneoic acid (DHA) in breast milk than in formula has been proposed as one explanation for this association and as a rationale for adding DHA to infant formula, but few long-term data support this possibility.
Objective: The objectives of the study were, first, to test the hypothesis that breastfeeding benefits stereoscopic visual maturation and, second, if that benefit is shown, to ascertain whether it is mediated by the dietary intake of DHA.
Design: Stereoacuity was measured by using the random dot E test (primary outcome), and visual acuity was measured by using the Sonksen-Silver acuity system (secondary outcome) in previously breastfed (n = 78) or formula-fed (n = 184) children aged 4–6 y who had been followed prospectively from birth. In the formula-fed group, children were randomly assigned to receive formula with either DHA or arachidonic acid (n = 94) or a control formula (n = 90) for the first 6 mo.
Results: Breastfed children had a significantly (P = 0.001) greater likelihood of foveal stereoacuity (high-grade or <100 s/arc) than did formula-fed children (odds ratio: 2.5; 95% CI: 1.4, 4.5) independent of potential confounding (P = 0.005). Stereoacuity did not differ significantly between children randomly assigned to DHA-supplemented or control formula. None of the groups differed in Sonksen-Silver visual acuity.
Conclusions: These findings support the hypothesis that breastfeeding benefits long-term stereoscopic development. An effect of DHA cannot be excluded, but the lack of difference in stereoacuity between infants randomly assigned to DHA-containing and those assigned to control formula raises the hypothesis that factors in breast milk other than DHA account for the observed benefits.
Key Words: Polyunsaturated fatty acids • stereoacuity • vision • breastfeeding • randomized trial
INTRODUCTION
Breastfeeding confers health benefits to infants, but whether it confers benefits beyond infancy has become a major focus of recent research (1). One area of controversy is the influence of breastfeeding on the visual development of healthy infants born at term. Several studies have shown greater visual acuity in breastfed than in formula-fed infants (2–9), although this finding is not consistent (10–19). Although breastfeeding was previously shown to be associated with higher stereoacuity at 3 y of age (2, 9), fewer studies have prospectively followed subjects beyond the first year of life (19).
A potential explanation for the inconsistent influence of breastfeeding on visual function is confounding by socioeconomic status (SES) and demographic factors and health behaviors that differ between mothers who breastfeed and those who formula-feed their infants. Such differences in sociobiological attributes may explain the accelerated visual development of breastfed infants, but relatively few studies have collected such data (9, 17–20). Alternatively, dietary constituents of breast milk that are absent from some formulas could contribute to the benefits of breastfeeding. Proposed constituents include the n–3 series of long-chain polyunsaturated fatty acids (LCPUFAs) and particularly docosahexaneoic acid [DHA; 22: 6n–3 (21, 22)]. DHA concentrations are higher in the brains (23, 24) and red cells (4–8, 11,13,15–17, 21, 22) of breastfed infants than in those of formula-fed infants, and DHA is particularly concentrated in retinal photoreceptor membranes (25). These observations, together with evidence from randomized trials showing short-term benefits for visual acuity with consumption of formulas that contain DHA (21, 22, 26), suggest that a higher DHA content may partly explain the benefits of breast milk for vision. However, few large trials with long-term follow-up exist to support or refute this view (19).
In a prospective follow-up of breastfed and formula-fed children who previously participated in a study of DHA supplementation (27), we tested the hypothesis that breastfeeding enhances stereoacuity in childhood independent of potential confounding factors. In addition, because formula-fed infants were randomly assigned to receive formula with and without DHA, we investigated whether the addition of DHA to formula affected stereopsis and thus whether any influence of breastfeeding was likely to be a consequence of the higher DHA content of breast milk. We chose stereoacuity as our principal outcome because stereopsis has been found to be better at age 3 y in infants whose mothers received a DHA-rich antenatal diet (9) or who were breastfed (2, 9). Therefore, the current study allowed us to assess whether supplementation of infant formula with DHA would be beneficial to long-term visual function.
SUBJECTS AND METHODS
Study population
Participants (n = 447) had previously participated in a randomized trial of DHA supplementation (27) or had agreed to be part of a breastfed reference group recruited alongside trial participants from 4 hospitals in Nottingham and Leicester, United Kingdom, between 1993 and 1995 (27). Participation was offered to the mothers of healthy singletons who were of appropriate size for their length of gestation, had a gestation time of 37 wk, and did not have any congenital abnormalities known to affect development. Breastfed infants were eligible if still breastfeeding at age 6 wk. For the formula-fed arm, only infants of mothers who had decided not to breastfeed and had begun formula feeding were recruited.
Written informed consent was obtained from all participants. The study was approved by the ethics committees of the collaborating centers and the Dunn Nutrition Unit, Medical Research Council, Cambridge, United Kingdom.
Study design
Maternal demographic, obstetric, and anthropometric data, including the number of cigarettes smoked per day during pregnancy, were collected at baseline. Socioeconomic status (SES) was based on the occupation of the parent who provided the main financial support for the family (or, if both parents worked, on the father's occupation) according to the classification of the Registrar General of the United Kingdom. For the initial phase of the study (27), children were seen at ages 6 and 12 wk and 6, 9, and 18 mo. At each visit, clinical and anthropometric data were collected. The main outcome measures during this part of the study were the Bayley Scales of Infant Development at 18 mo of age (27).
Formula-fed infants were randomly assigned during the first week of the study to either a standard or DHA- and arachidonic acid (AA)–supplemented formula (27). A random permuted block design, stratified by center (Nottingham or Leicester) and sex, was used; assignments were concealed with the use of sealed opaque envelopes (27). Infants were provided with the assigned formula until age 6 mo.
Study formulas
The nutritional compositions of the 2 trial formulas are presented in Table 1. The formulas (Nestec Ltd, Vevey, Switzerland) were prepared in powdered form for the intervention period (liquid formulas were used only at the hospital). The formulas were the same except for the manipulation of fat sources designed to provide 0.32% DHA and 0.30% AA in the total lipid of the LCPUFA-supplemented formula and some alterations in the contents of cholesterol and other fatty acids resulting from the addition of DHA and AA. It was intended that the sole source of LCPUFA would be purified egg phospholipid (Lipid Teknic, Stockholm, Sweden), and 8 of 12 product batch-runs achieved that goal. Quality control showed that the DHA content was below target in 2 of 12 batches, and the shortfall was made up by the addition of a small amount of high-DHA fish oil (5% of total LCPUFAs). In addition, it was necessary to use an alternative source of lipids as egg phospholipids (Lucas-Meyer, Hamburg, Germany), which had 70% phospholipids, again supplemented with a small amount of high-DHA fish oil (5% of total). Thus, DHA was mainly supplied from egg lipids, predominantly phospholipid. The total phospholipid content of the standard and LCPUFA-supplemented formulas was 0.5 and 2.2–2.6 g/L formula, respectively. The LCPUFA sources represented 7–11% of the total fat blend. The formulas were identical in appearance and were identified only by codes, which were not known to any of the study personnel or participants.
View this table:
TABLE 1. Composition of infant formulas
Follow-up
Children from all families who were traceable and who agreed to participate in this further follow-up were reviewed between the ages of 4 and 6 y. Assessments of cognitive function (to be reported elsewhere) and visual development were conducted in the child's home or, on occasion, at the school by 1 of 3 observers. The follow-up of subjects at 9 and 18 mo and 4–6 y is shown in Figure 1.
FIGURE 1.. Progress of the trial to subject ages 4–6 y. *Subject could not be traced or had left the country.
Visual assessment
Vision was assessed by an examiner who was either a doctor or a psychologist and who was specifically trained in the assessments required. After a standard examination to check for eye abnormalities and strabismus, the examiner measured stereoacuity by using the random dot E test (primary visual outcome) (28–31) and measured visual acuity by using the Sonksen-Silver acuity system (secondary visual outcome) (32, 33). Stereoacuity was chosen as an outcome because it is a cortical phenomenon that is expected to reflect differences between persons in the maturity of the visual cortex (9) and that has been suggested to be influenced by infant nutrition (2, 9). The random dot E test was chosen because it is a reliable and easily administered screening test of stereoacuity in children of this age group (29) and because 2 previous studies showed an advantage for breastfed over formula-fed infants with respect to steroacuity by using a similar test (2, 9). The Sonksen-Silver acuity system was used because the displays are in accord with the Snellen International Standard for crowded linear displays used routinely in adults and therefore are meaningful in terms of the adult standard (32, 33). In addition, it is the only standard test that provides age norms.
In accordance with the standard format for the administration of the random dot E test, the child put on 3-dimensional viewing glasses, was shown a physical model of the letter "E," and was asked to point to the test card (1 of 2) showing a similar character. Once the child had understood the procedure, the random dot E test card and the stereo blank were presented at a distance of50 cm from the subject, and the child was again asked to point to the card with the letter "E." Next, the test and stereo blank cards were shuffled behind the tester's back, and the choice was presented again at a distance of 50 cm. The test was repeated at distances increasing by 50 cm from the child. Four consecutive correct responses at each distance were taken as a pass at that distance.
The Sonksen-Silver acuity system was administered as specified in the manual. The tester first familiarized the child with the keycard by pointing out each letter in turn. The child was then asked to match each single letter in the training booklet to the keycard. The tester then presented the test booklet to the child at a distance of 3 m. Initially, the tester ascertained the child's approximate visual level by pointing to the letter at the righthand end of the line in the size 18 m (ie, 18 min/arc) display and of successively smaller displays until the child failed to identify one letter. The tester then returned to the previous (larger) display and tested each of 5 letters. The measure of acuity was taken as the smallest display from which all 5 letters were identified. After measurement of binocular acuity, monocular acuity was measured by repeating the test while occluding each eye in turn.
Statistical analysis
Student's t test for continuous variables and chi-square analysis for dichotomous variables were used to compare children reviewed at age 4–6 y with those not followed. Stereoacuity data (primary outcome measure) were analyzed both as a dichotomous variable—defined as foveal (high-grade or adult, ie, <100 s/arc) or nonfoveal, as described (9)—and as a continuous variable. For analyses as a continuous variable, stereoacuity data were loge transformed and then multiplied by 100 (34). The SD for the transformed data represents the CV (%), and regression coefficients represent the percentage change in stereoacuity per unit change in the independent variable (34).
The 3 dietary groups (breastfed and control and LCPUFA-supplemented formulas) were compared by using analysis of variance and Tukey's test for post hoc comparisons of continuous variables and chi-square analysis for dichotomous variables. In secondary analyses, stereoacuity was compared in breastfed and formula-fed children (control and LCPUFA-supplemented formula-fed groups were combined) and adjusted for potential confounding factors by using multiple linear and logistic regression analysis for continuous and dichotomous outcome variables, respectively. Stereoacuity and visual acuity were also compared in infants randomly assigned to receive formula with and without DHA supplementation. All analyses were conducted by using SPSS for WINDOWS software (version 12.0; SPSS Inc, Chicago, IL).
RESULTS
Study sample
Of the 138 breastfed and 309 formula-fed infants recruited at birth, a subsample of 78 (57%) breastfed and 184 (60%) formula-fed children was evaluated at age 4–6 y (Figure 1). This sample size was sufficient for detection, with 80% power at 5% significance, of a difference of >0.4 SD (9%) in stereoacuity between breastfed and formula-fed groups and between DHA-supplemented and nonsupplemented groups.
The children from both breastfed and formula-fed groups who were followed were representative of those recruited at birth in terms of their sex ratio, gestation, birth weight, length, and head circumference (Table 2). There was no evidence of a statistical interaction between type of feeding and follow-up status with respect to maternal, anthropometric, socioeconomic, and demographic characteristics at birth. At 4–6 y of age, the 3 dietary groups were well balanced for demographic and anthropometric characteristics at birth (Table 3), but breastfed infants differed significantly from those fed control or LCPUFA-supplemented formula in their parental anthropometry and in the mother's age, SES, and level of education (Table 3).
View this table:
TABLE 2. Characteristics at enrollment of children followed and not followed at ages 4–6 y1
View this table:
TABLE 3. Characteristics of population studied1
Visual function
Of the 262 children followed, 6 had strabismus and another 13 did not complete the assessment of stereoacuity. Children who were breastfed were more likely to achieve foveal stereoacuity and had better stereoscopic vision (lower mean value for stereoacuity) than did those who were fed control or LCPUFA-supplemented formula (Table 4). No evidence was found of an interaction between breastfeeding and study center (P = 0.1), sex (P = 0.5), or SES (P = 0.9) with respect to stereoacuity at age 4–6 y.
View this table:
TABLE 4. Visual acuity at ages 4–6 y1
In secondary analyses, breastfed children were significantly more likely to achieve foveal stereoacuity (62%, 45/78) than were those fed formula (control and LCPUFA-supplemented formula groups combined: 39%, 66/184); the values were OR: 2.5; 95% CI: 1.4, 4.5; P = 0.001. This association remained after adjustment for SES (adjusted OR: 2.5; 95% CI: 1.3, 4.6; P = 0.005), maternal age (adjusted OR: 2.6; 95% CI: 1.4, 4.7; P = 0.002), and maternal education (adjusted OR: 2.5; 95% CI: 1.3, 4.8; P = 0.004) or for factors previously shown to be associated with visual function—ie, parental SES and cigarette smoking, age at testing, and head circumference (18), as well as birth weight and length (adjusted OR: 2.5; 95% CI: 1.3, 4.6; P = 0.005). Mean stereoacuity in breastfed children (97 s/arc; CV 24%) was significantly higher than that in those fed formula (109 s/arc; CV 25%; n = 184); the mean difference was –11% (95% CI for difference: –4, –18%; P = 0.002). This difference remained after adjustment for SES (mean difference: –11%; P = 0.005), maternal age (mean difference: –12%; P = 0.001), maternal education level (mean difference: –12%; P = 0.003) or for potential confounding factors, as above (mean difference: –11%; P = 0.004).
No evidence was found that stereoacuity differed significantly between children fed control or LCPUFA-supplemented formula, as measured by the likelihood of having foveal stereoacuity (OR: 1.2; 95% CI: 0.6, 2.1; P = 0.6; mean difference: –0.2%; 95% CI: –7, 7%; P = 0.9), even after adjustment for potential confounding factors, as above (adjusted OR: 1.2; P = 0.6; mean difference: –0.8%; 95% CI: –8, 7%; P = 0.8). The proportion of children with a visual acuity of 3/3 measured by the Sonksen-Silver system did not differ between the 3 dietary groups (Table 4) or between those fed an LCPUFA-supplemented or a control formula (P = 0.9 for comparisons of left or right eye or both eyes).
Socioeconomic and biological correlates with stereoacuity
In linear regression analyses, no evidence was found that factors other than breastfeeding were associated with random dot E test stereoacuity at ages 4–6 y (data not shown). These factors included SES; maternal cigarette smoking, age, height, and educational level; and the child's gestation, head circumference, birth weight or length, and age at testing.
DISCUSSION
We found that breastfed children had greater random dot E test stereoacuity than did those fed formula, independent of potentially confounding factors. Indeed, no factor that we identified except breastfeeding was associated with stereoacuity. Our findings in children whose visual function is approaching maturity therefore support the hypothesis that breastfeeding benefits stereoscopic visual development. It is important, however, that, whereas benefits in visual development have been hypothesized to relate to the presence of DHA in breast milk—this is part of the rationale for the addition of DHA to infant formulas—we found no difference in stereoacuity between children randomly assigned to receive formula with and without DHA. Although we recognize some limitations of our study (as discussed below) and realize that the source of DHA used in formulas may be influential, the current study raises the possibility that factors other than the high DHA content of breast milk could account for, or at least contribute to, the benefits seen in breastfed infants.
The evidence for an influence of early diet on visual function has been inconsistent, possibly because of the different electrophysiological or behavioral tests used, different ages of the children at follow-up, and the small numbers of children studied. In contrast with most previous reports, the current study followed children to an age at which standard tests of visual acuity could be used, and it was powered to detect small differences between dietary groups. Our principal outcome, stereoacuity as measured by the random dot E test, is a cortical phenomenon and may better reflect the maturity of the visual system in young children than does visual acuity (9). Stereoacuity measured with this test was in the range of 100 to 120 s/arc, which is consistent with standards for children of a similar age but below that expected for adults (70 s/arc) (30). Therefore, as stereoscopic visual function is not fully mature in the age range studied (30), our findings suggest that breastfeeding is associated with faster stereoscopic visual development. Consistent with this hypothesis, breastfed infants were more than twice as likely (OR: 2.5) to have high-grade or adult stereoacuity than were formula-fed infants; this effect size is similar to that in a previous study showing that breastfeeding (as well as a higher DHA consumption in pregnancy) was associated with greater stereoacuity in slightly younger children (9). However, further follow-up is needed to resolve whether stereoacuity in formula-fed infants catches up to that in breastfed infants as they age beyond early childhood. In contrast with stereopsis, measurements of visual acuity by using the Sonksen-Silver acuity system may have lacked the sensitivity to pick up small potential differences between dietary groups, because most children (67%) will have perfect visual acuity (monocular acuity of 3/3) at age 5 y, and their acuity will rise to 89% by age 6.5 y (33).
Our findings confirm the findings from previous reports, ie, that stereoacuity is more mature in breastfed than in formula-fed children (2, 9). This difference was seen even after adjustment for potential confounding factors. Furthermore, stereoacuity did not correlate with factors previously associated with visual development, such as maternal age (9), parental SES or cigarette smoking, or the child's anthropometric values at birth (17, 18, 20). Although the comparison between breastfed and formula-fed children was not randomized and therefore cannot prove causation, our findings suggest that the influence of breastfeeding on later stereoacuity is not related to population differences between children who are breastfed or formula-fed. Therefore, although unidentified confounding could exist, and a follow-up rate of 60% could have introduced confounding as a result of nonrandom loss, our data support the hypothesis that breastfeeding itself rather than its sociobiological correlates benefits stereoscopic visual development.
The associations of maternal fish consumption during pregnancy with later stereoacuity (9) and of red cell DHA concentrations with later visual function have provided support for the hypothesis that DHA in breast milk accounts for the effect of breastmilk on stereoacuity (3–5, 35). Dietary intake of DHA could have differed between women who chose to breastfeed rather than formula feed, and this difference could have contributed to the visual advantage of breastfed infants. However, DHA concentrations may also correlate with other markers of a healthy lifestyle. In our experimental study, stereoacuity did not differ significantly between children randomly assigned to a DHA-supplemented or control formula, which raises the possibility that the advantage of breastfeeding is related, at least in part, to factors other than the higher DHA content of breast milk than of formula.
Our findings have several potential explanations. First, DHA supplementation may have only a transient benefit for visual function in infancy (6, 21, 26). First, our findings were consistent with those of a recent smaller study, which showed that LCPUFA-supplemented formula improved stereoacuity in infants aged 17 wk but not in those aged 52 wk (36). Second, the amount of DHA supplementation may have been inadequate to influence visual function, although that is unlikely, because the DHA content of our supplemented formula (0.3% of fatty acids) is similar to that of breast milk in many populations, and it is above recently recommended concentrations (0.2%) (37). Nonetheless, despite adequate supplementation, DHA concentrations still could be lower in cellular membranes of formula-fed than of breastfed infants and thus could be insufficient to benefit stereoscopic visual maturation. We chose not to carry out a venipuncture (although that decision could potentially limit the study) because that procedure could impair compliance at later follow-ups and because the red cell LCPUFA concentration does not necessarily correlate with function (11, 13, 15). The lack of data on red cell LCPUFA concentrations also means that we cannot be certain of compliance, although significant noncompliance was unlikely, given a study population fed exclusively on formula given by us and monitored carefully throughout the supplementation period (27).
Third, the lack of benefit of DHA supplementation on later stereoacuity may result from the fact that the DHA in the supplemented formula was derived from egg phospholipids, rather than from triacylglycerols (its source in human milk). Theoretically, DHA from a triacylglycerol source may be required for optimal brain development in infants, although, to our knowledge, no data strongly support or refute this hypothesis. Fourth, a type II error could also explain our findings, although that possibility was unlikely because the current study was powered to detect a difference of 0.4 SDs between randomly assigned, formula-fed groups—a difference smaller than the difference in stereoacuity that we found between breastfed and formula-fed children. Nevertheless, our trial may have missed smaller but clinically significant differences in stereoacuity between DHA-supplemented and control groups. For instance, by reducing the variability in stereoacuity associated with age, a narrower age window at examination could have helped detect smaller differences between groups. Finally, limitations of the random dot E screening test, which is sensitive to minor refractive errors—particularly if refraction is unequal in the 2 eyes—may have also restricted our ability to pick up small differences between infants fed DHA-supplemented and those fed control formula, although the randomized study design should have prevented any systematic bias.
We did not find a beneficial effect of DHA supplementation of formula on long-term stereoacuity, but the marked difference in stereopsis between breastfed and formula-fed children requires explanation. One possibility is that the act of breastfeeding itself stimulates the maturation of visual function, although the lack of any association between maternal education or SES (both of which tend to correlate with the degree of maternal stimulation) and stereoacuity suggests that other mechanisms may also contribute. Another possibility is that, whereas red cell DHA concentrations reflect the extent of breastfeeding, factors in breast milk other than DHA promoted visual maturation. Either nonnutritive factors or dietary factors present in breast milk but absent from formulas or an influence of breastfeeding on concentrations of hormones such as thyroxine could be responsible for the maturation (1). Theoretically, factors in formula could also have a negative effect on visual development, although no evidence exists to support this possibility.
Overall, our study adds to the increasing evidence that breastfeeding benefits stereoscopic visual development. We postulate that this advantage could contribute to the positive association between breastfeeding and later cognitive development. Further research into the mechanisms by which breastfeeding benefits stereoacuity is therefore warranted.
ACKNOWLEDGMENTS
We thank all the mothers and children who took part in this study. We also thank Geraldine McHugh, Caroline Browne, and Claire Lawson for carrying out the visual assessments and Emma Sutton for tracing the study participants.
AS was the chief investigator, supervised data collection and wrote the first draft of the manuscript. RM and AL initiated the original study. KK, EI, MF, AE-J, and TS contributed to study design. TJC provided statistical expertise, and PS provided expertise in visual assessment. All authors contributed to the final draft. None of the authors had a personal or financial conflict of interest.
REFERENCES